Skip to main content

Full text of "Identification of Reactive Power Reserve in Transmission Network"

See other formats


ACEEE Int. J. on Control System and Instrumentation, Vol. 02, No. 01, Feb 2011 

Identification of Reactive Power Reserve in 
Transmission Network 

Robert A. Lis 1 , Grzegorz Blajszczak 2 , and Magdalena Wasiluk-Hassa 2 

1 Wroclaw University of Technology/ W5 -18, Wroclaw, Poland 

Email: robert.lis@pwr.wroc.pl 

2 Polish Transmission System Operator, Warsaw, Poland 

Email: {grzegorz. blajszczak magdalena. wasiluk-hassa} @pse-operator.pl 



Abstract — The paper describes importance of the reactive 
power control basing on the failures and control problems in 
the Polish transmission networks. A detailed description of the 
operational difficulties is provided. The paper also presents a 
highly automated method identifying voltage control areas 
(VCA), areas prone to voltage instability, and reactive power 
reserves requirement ensuring voltage stability under all 
considered contingencies. During the completion of the VCA 
project testing of the VCA software was limited to power flow 
models of the Polish Power Grid Operator (PSE) System. A 
detailed description of the operational difficulties is provided. 
Conclusions, repairs and prevention undertaking are also 
described. 

Index Terms — Power System Stability, Voltage Security 
Assessment, Voltage Stability, Intelligent Systems. 

I. Introduction 

The quality of the electrical energy supply can be 
evaluated basing on a number of parameters. However, the 
most important will be always the presence of electrical 
energy and the number and duration of interrupts. If there is 
no voltage in the socket nobody will care about harmonics, 
sags or surges. A long term, wide-spread interrupt - a 
blackout leads usually to catastrophic losses. It is difficult to 
imagine that in all the country there is no electrical supply. 
In reality such things have already happened a number of 
times. One of the reason leading to a blackout is reactive 
power that went out of the control. When consumption of 
electrical energy is high, the demand on inductive reactive 
power increases usually at the same proportion. In this 
moment, the transmission lines (that are well loaded) 
introduce an extra inductive reactive power. 

The local sources of capacitive reactive power 
become insufficient. It is necessary to deliver more of the 
reactive power from generators in power plants. It might 
happen that they are already fully loaded and the reactive 
power will have to be delivered from more distant places or 
from abroad. Transmission of reactive power will load more 
the lines, which in tern will introduce more reactive power. 
The voltage on customer side will decrease further. Local 
control of voltage by means of autotransformers will lead to 
increase of current (to get the same power) and this in tern 
will increase voltage drops in lines. In one moment this 
process can go like avalanche reducing voltage to zero. In 
mean time most of the generators in power plants will switch 
off due to unacceptably low voltage what of course will 

©2011 ACEEE 

DOI: 01.IJCSI.02.01.52 



deteriorate the situation. In continental Europe most of the 
power plant are based on heat and steam turbines. If a 
generation unit in such power plant is stopped and cool down 
it requires time and electrical energy to start operation again. 
If the other power plants are also off - the blackout is 
permanent [1]. 

II. An Operational Difficulties In Transmission 
Networks Arose From Reactive Power 

The difficulties showed up on June 26, 2006. The 
prediction for power consumption on this day was 18200 
MW (in the morning peak) what was much higher compared 
with June in last year or previous years. This power was 
planed to be supplied from 75 generation units. Above these, 
there were a hot power reserve of 1350 MW (in this 237 
MW second-reserve, 656 MW minute-reserve) and a cold 
reserve of about 2600 MW. In the north-east Poland there is 
not any grid-generation. The closest to this region is 
Ostroleka Power Plant, which in that time from three 200 
MW units has two in operation and one set off for 
maintenance. In early morning of the 26th one unit in Power 
Plant Patnow had to be switched off and before noon four 
other units (two in Kozienice P. P. and two in Laziska P. P.) 
were switched off as well. All these unites were the main 
supplier to the north-east region of Poland. At 7 o'clock 570 
MW of power was lost. At the same time the consumption 
prediction appeared to be wrong - the consumption was 600 
MW higher and there was also much higher demand on 
reactive power. At 13 o'clock there was an unbalance of 1100 
MW. In mean time one unit (in Dolna Odra P. P.) had been 
activated. However further activation from cold-reserve 
required more time (about 6 hours) because of technological 
reasons. Unusual heat wave spreading throughout the country 
caused deterioration of the operational conditions in power 
plants. Due to lack of sufficient amount of cooling water 
and exceeded water temperature levels the generating 
capacities of some power plants systematically decreased. 
That situation concerned mainly the power plants located in 
the central and northern part of Poland, the loadings of some 
transmission lines reached the acceptable limits what in turn 
cause the necessity of generation decrease in power plants 
located outside the mentioned region. The control of reactive 
power became critical. About noon the voltages were low, 
but still within limits. Rising demand and lack of additional 
reactive power sources brought further voltage decrease. 



20 



^c ACEEE 



ACEEE Int. J. on Control System and Instrumentation, Vol. 02, No. 01, Feb 2011 



The transformer voltage control had as a priority to keep 
constant the voltages in 110 kV networks. At 13 hours, most 
voltages in central and north Poland were below the limits. 
The generation units in Ostroleka P. P. worked with full power 
providing also about 100 MVA of reactive power each. They 
worked with automatic control of reactive power generation. 
Further increase on reactive power demand caused power 
oscillation between these two units, and as a consequence 
switching off one of them (due to large current) at 13:04. 
The voltage went down and in four minutes the second unit 
was also off because the voltage was too low. Lost of 400 
MW and 200 MVA reactive in critical region affect 
dramatically all the system. The voltage became much below 
acceptable limits. All small, local power stations and heat- 
combine power stations were off immediately. A big unit in 
Kozienice P. P. was off at 13:08. A DC link to Sweden, that 
supplied the region with 300 MW was off, as well as its 
huge battery of capacitors. To rescue the power grid 
additional power from neighbouring countries has been 
bought: 400 MW from Czech Rep., 100 MW from Slovak 
Rep. and 500 MW from Germany. In this time a sharp 
reduction of consumption was introduced. Switching off 
about 100 MW loads and a few lines allowed to stabilized 
the voltages in north of Poland and to put back in operation 
the units in Ostroleka and Kozienice P. P. About 16 hours 
the system operation returned to normal conditions. 

III. A Reactive Power Management In Transmission 
Networks 

The crises situation described above shows a 
development of blackout without any unusual events like 
explosion, storm, hurricane, etc. The existing generation was 
sufficient to cover all the demand. The transmission capacity 
was much higher than necessary. The power grid is equip 
with automatic reactive power and voltage control system. 
This system also worked correctly. Despite the above a 
serious problem appeared. The reactive power compensation 
is a service that is location dependent. Transmission of 
reactive power over long distances is not only not economical 
but also ineffective from control point of view. Reactive 
power flow involves generation of additional reactive power. 
In normal operating conditions the control system manage 
to adjust generation of reactive power in power plants and 
to control the voltages. If the location of reactive power 
sources is inadequate, the control system may lead the power 
grid to a blackout [2] . 

Since it is well understood that voltage security is 
driven by the balance of reactive power in a system, it is of 
particular interest to find out what areas in a system may 
suffer from reactive power deficiencies under some 
conditions. If those areas prone to voltage security problems, 
often called critical Voltage Control Areas (VCA), can be 
identified, then the reactive power reserve requirements for 
them can also be established to ensure system secure 
operation under all conditions [3]. To identify VCAs in a 
given power system, the considered system is stressed to its 
stability limit for various system conditions under all credible 



contingencies. At the point of instability (nose of the PV 
curve) modal analysis is performed to determine the critical 
mode of voltage instability for which a set of bus participation 
factors (PF) corresponding to the zero eigenvalue (bifurcation 
point) is calculated. Based on these PFs, sets of buses and 
generators that form the various VCAs in a given power 
system are identified. The identification procedure applies 
heuristic rules to (a) group contingencies that are related to 
the same VCA; and (b) identifies the specific buses and 
generators that form each VCA as described below. The 
identification program processes the sets of buses and 
generators corresponding to the PFs obtained from the modal 
analysis for each system condition and contingency case. 
Contingencies are clustered if their sets of bus PFs are similar. 
Finally, the program identifies the sets of buses and 
generators that are common to all contingencies of each 
cluster. Those sets of buses and generators form the VCAs 
of the power system. 

The VS AT program is used to simulate the scenarios 
and to compute PV curves for all transfers and contingencies. 
The objective is to stress the system in the manner specified 
by the given transfer and to perform modal analysis at the 
nose point of the PV curve. Modal analysis outputs include 
the critical mode eigenvalue (zero at the PV nose point), 
critical mode bus participation factors, and generators that 
are at their reactive power limit. All generated output files 
are collected for post-processing in order to generate the 
database (DB) records for the VCA identification engine. 
Each VCA identified is related to a cluster of contingencies; 
these cases are said to "support" that VCA. First, similar 
contingency cases are clustered and then second, the specific 
buses and generators that form the VCAs are identified. 
Before clustering contingency cases, however, a preliminary 
selection of buses and generators is done at an earlier stage 
of the VCA identification process as shown in Figure l.The 
VCA identification process consists of the following steps 
[3]: 

1) Selection of Buses for VCA Identification - From 
modal analysis results for each contingency, a subset of buses 
with high PF is selected for further analysis (SFAs): remaining 
buses are discarded. Several strategies to select such subset can 
be applied. Generator terminal buses appear in the PF only if 
the generator exhausts its reactive power reserves (marked as a 
Q-limited, QLbus). 

2) Clustering of Contingency Cases based on SFAs - the 
identification program clusters contingency cases based on 
similarities. These clusters will be used to identify the VCAs in 
the power system, as described in Steps 6 and 7 below 

3) Normalization of Generator Buses PFs - the generator 
buses PFs are normalized. 

4) Selection of Generators in Cluster Ck - For each cluster 
Ck, the frequency of generator bus participations in this Ck is 
calculated. The generator buses with the highest frequencies 
are selected to represent the cluster Ck reactive power reserves 
and are denoted as GENk. 

5) Clustering of Ck based on GENs - In this step, a number 
of Ck are grouped together if their corresponding GEN sets are 



©2011 ACEEE 

DOI: 01.IJCSI.02.01.52 



21 



h ACEEE 



ACEEE Int. J. on Control System and Instrumentation, Vol. 02, No. 01, Feb 2011 



similar. Two GENs are considered similar if certain percentage 
of generator buses are matched. If GENi (from Ci) and GENj 
(from Cj ) are similar, then Ci and Cj are grouped together into 
a preliminary VCAm. This VCAm is associated with a set of 
generator buses GENm that consists of the generator buses of 
the combined GENi and GENj. The first step in clustering Ck 
is to select the base set GENx to which other GEN's are 
compared. 

6) VCA identification part A: Selection of buses - For each 
preliminary VCAm, compute the frequency of each bus. Then 
select the buses with a frequency greater than a user defined 
threshold value. 

7) VCA identification part B: Selection of generators - For 
each GENm, get the frequency of each generator bus. Then 
select the generator buses with a frequency greater than a user 

Si..-|- i 



defined threshold value The generators associated with these 
generator buses are the ones that form controlling generators 
associated with VCAm 

A. Heuristic Rules for Base Selection and Similarity 
Measurement 

Selection of a base for clustering process - From 
the VCA identification process presented in the previous 
section we can observe that clustering is carried out twice: 
Clustering contingency cases as described in Step 2 and 
Clustering Ck based on GENs as described in Step 5. 
Measure of similarity between buses/generators sets - The 
number of common elements C is counted and compared 
with a similarity of a user defined threshold T. If the number 
of common elements C is greater than the threshold T, then 




CluMOriTi.c 
L*onLiii£L'[u. , ses 




<p Step 2 



SFAi=(b*ises.EHHieriitar-lMLEes] jf 
C-XH5FA-L 5FA-J, ...} 



VCAs in 
Power System 



Identification Specific 
buses iltkI £,ens in VCAs 











C-l ' 


Selection of 
Gene mi ore 


^ i 




^ l 




GEN-s 




C-X I J 


Selection or 
Generator* 

*- 






/ 


. 


i 
1 

o 

m 
-z. 


S I c p 4 






r \, 
Clustering C-s 
using GEN-s 











Steps 6 and 7 
Figure 1 . VCA Identification Process 



S,: V ? 



set-/ and the base set are considered being similar. The 
similarity threshold T is set as a percentage of the number of 
elements in the largest set (set-/ or the base set). 

B. Reactive Power Reserve Requirements 

After the VCAs have been identified, it is desirable 
to know what reactive power reserves are necessary to 
maintain in the system in order to ensure voltage stability 
under all conditions. In the section above describing VCA 
determination, the generators that control each VCA were 
identified. It is for these generators that reactive reserve 
requirement must be established for each VCA. For each 
scenario analyzed, the pre-contingency reactive output on 
each controlling generator is recorded at a "secure point" 
back from the nose of the PV curves. The "distance back 
from the nose" corresponding to Point "Post Contingency 
stability limit" is determined based on the required security 
margin criteria (such as 5% of the transfer load or generation). 



IV Summary 

An overview of the state-of-the-art of on-line VS A has been 
presented and a highly automated method for the 
identification of voltage control areas is described. Voltage 
control areas describe the regions in a power system that 
under specific conditions are prone to voltage instability. 
Intelligent systems hold promise to improve VSA speed, 
provide adaptive learning capabilities and offer the ability 
to identify key system parameters. An example of an 
intelligent system framework using decision trees has been 
described. Work in this area is continuing toward a pilot 
implementation at a PSE Operator host site. 

References 

[1] Definitions and Classification of Power System Stability" 
IEEE/OGRE Joint Task Force on Stability Terms and Definitions, 
IEEE Transaction s on Power Systems Vol. 19, No. 2., May 2004. 
[2] CIGRE Technical Brochure on Review of On-Line Dynamic 
Security Assessment Tools and Techniques, CIGRE Working Group 
C4.601, June 2007. 

[3]EPRI Final Report 1013995, "Identification of Critical Voltage 
Control Areas and Determination of Required 
ReactivePowerReserves",October,2004. 



©2011 ACEEE 

DOI: 01.IJCSI.02.01.52 



22 



^c ACEEE